Blackwell entry-level GPU skimps on bandwidth
Nvidia’s latest budget card, the GeForce RTX 5050, is shaping up as another exercise in doing just enough.
Apple defends its headless chicken AI strategy
Job’s Mob reckons building Siri properly takes decades
After months of mocking silence about Siri’s “smarts” Apple executives have surfaced to defend their wobbly AI rollout and somehow made it sound like all part of the plan.
AI firms train abroad to bypass export controls
Chinese firms are bypassing US chip bans not by smuggling silicon but by flying training data overseas and using rented Nvidia-powered servers in Malaysia to do the heavy lifting.
AI firms race to fix sycophantic chatbots
Before users get too comfortable
OpenAI, Google DeepMind and Anthropic are working to rein in a growing problem with their chatbots: excessive flattery. The models are increasingly prone to giving users agreeable responses that prioritise validation over accuracy.
PCIe 7.0 spec screams ahead of reality
Blistering speeds for kit you won’t own this decade
PCIe connectivity is moving ahead at full tilt, with a new spec already outlined for PCIe 7.0. Trouble is, we’re still mostly faffing about with PCIe 5.0, and PCIe 6.0 is barely out of the gate.